My wife pointed me toward this PhysOrg report on a study by Dr. Dhriti Sengupta and Dr. Ananyo Choudhury in the Sydney Brenner Institute for Molecular Bioscience (SBIMB) at Wits University:
A new study challenges the presumption that all South-Eastern-Bantu speaking groups are a single genetic entity. The South-Eastern-Bantu (SEB) language family includes isiZulu, isiXhosa, siSwati, Xitsonga, Tshivenda, Sepedi, Sesotho and Setswana. Almost 80% of South Africans speak one of the SEB family languages as their first language. Their origins can be traced to farmers of West-Central Africa whose descendants over the past two millennia spread south of the equator and finally into Southern Africa.
Since then, varying degrees of sedentism [the practice of living in one place for a long time], population movements and interaction with Khoe and San communities, as well as people speaking other SEB languages, ultimately generated what are today distinct Southern African languages such as isiZulu, isiXhosa and Sesotho. Despite these linguistic differences, these groups are treated mostly as a single group in genetic studies. […]
“South Eastern Bantu-speakers have a clear linguistic division—they speak more than nine distinct languages—and their geography is clear: some of the groups are found more frequently in the north, some in central, and some in southern Africa. Yet despite these characteristics, the SEB groups have so far been treated as a single genetic entity,” says Choudhury. The study found that SEB speaking groups are too different to be treated as a single genetic unit. “So if you are treating say, Tsonga and Xhosa, as the same population—as was often done until now—you might get a completely wrong gene implicated for a disease,” says Sengupta.
More at the link, of course; I’m curious what the Hattic genetics experts make of this.
How are you feeling today?
I was clicking idly through to other linked papers, including
https://phys.org/news/2017-05-genetic-analysis-reveals-patterns-migration.html
and found this remarkable statement in the summary there:
The researchers also obtained genetic information for approximately 5,200 African Americans and used it to investigate their African background. They report that they found a high degree of BSP [“Bantu speaking peoples”] ancestry, which was expected, as slave traders typically brought slaves to the Americas from the western coast of Africa. But the researchers report that they also found that their genomes on average were 16 percent non-BSP, suggesting that African Americans are much more genetically diverse than has been thought.
16% actually seems astonishingly low; most slaves came from West Africa and thus had no Bantu-speaking forebears at all. The figure isn’t in the abstract of the actual paper (which I don’t have access to), so it is (one hopes) the reviewer’s error rather than a sign of serious problems with the actual paper (like not knowing what “Bantu” actually means.)
The original paper is open access. Sengupta et al., 2021, Genetic substructure and complex demographic history of South African Bantu speakers, Nature Communications 12:2080.
(It’s an “article number”. That’s how they do it with these newfangled online publications.)
I was hoping for a post I could hitch this recent paper as an aside to: Idiatov and van de Velde, 2021, The lexical distribution of labial-velar stops is a window into the linguistic prehistory of Northern Sub-Saharan Africa. Language 97:72. “It shows that LV stops are a substrate feature that should not be reconstructed into the early stages of the languages that currently have them,” among other things.
Interesting that I saw it first over here, and that’s despite my main line of work being focused on expanding the successes of medical genetics to the patients with African ancestry. But the passage about “completely wrong” results is an exaggeration. Basically they showed that in a hypothetical study where the cases are from one ethnic group, but the controls are from another, then you can observe a bunch of artifacts. However there are standard good practices of controlling for such differences between the study cohorts, and emerging methods for actually extracting the valuable information even when the study cohorts are ethnically diverse and historically recently admixed. Including some methods I work on. Perhaps a bigger issue is that of a socio-economical confounding factors correlated with genetic ancestry (most commonly, depending on who one’s ancestors were, one may have different level of access to health care, and therefore enter a study with different probability depending on the severity of the trait … so instead of measuring who is more sick, you end up measuring, partly, who’s more disadvantaged). And an even bigger issue is that the disadvantaged minorities have far too little health and genetic data collected in the first place, both because they are disadvantaged and because they are numerically smaller than the best-studied populations of European descent). And I can’t believe that I am writing this stuff HERE.
As to the actual historical demographic results, they are hardly new IMHO. The differential intermixing with the Khoe San peoples is well known to both the linguists and the geneticists. The dates of this interaction are of little surprise too. The authors may have been the first to demonstrate female bias of this genetic admixture, but it’s almost always the same anywhere around the globe, with the in-migrating group receiving a female-biased genetic inflow … so no surprise here either. They also show detectable genetic differences between the successive waves of Bantu migrations even after accounting for Khoe-San DNA, but it’s probably unsurprising too, given that these waves picked some non-Bantu genetic components as they traversed East Africa, even before encountering the Khoe-San?
@Y:
Thanks for the Eddyshaw-bait paper!
Proto-Oti-Volta definitely had labiovelar stops, as do the vast majority of modern Oti-Volta languages (Mooré, coincidentally by far the largest, has secondarily reduced them to velars, however.) I note that this entire area (in fact, the whole “Gur” area) is basically a “here be dragons” lacuna on their nice maps. Manessy reconstructed labiovelars for his Proto-Gur; I don’t believe in Gur as a valid node, but Manessy made a very good case for projecting them back to before Proto-Oti-Volta, at any rate.)
Kusaal, which is pretty typical within Oti-Volta in this regard, has labiovelars in undoubted core vocabulary (e.g “die”, “skin”, “near”, “sleep”, “here”, “seize”), but actually not particularly in ideophones. Just off the top of my head, “moon”, “rope” and “calabash” all began with a nasal labiovelar in Proto-Oti-Volta (and still do in Mampruli, Dagbani, Nawdm, Gurmanche …)
Fancy maths is worthless when based on crap data.
The more competent/less evil journals precede it by the letter e to make sure everyone understands it’s not a page number…
I figured it might be Eddyshaw-bait, but didn’t know how much it is so. I’m generally suspicious of “there’s too much forest to bother looking at the trees” papers, which the majority of linguistic phylogenetics studies are.
But now, to make peace, could some language akin to Oti-Volta have been the substrate which provided the Bantu languages with LVs?
16% actually seems astonishingly low; most slaves came from West Africa and thus had no Bantu-speaking forebears at all.
No, they actually MEAN 16%, specifically in those ancestors who came from Angola and the surrounding area. The average in all African ancestors of the African Americans was estimated even lower, at 4.8%, but you shouldn’t conflate their “BSP ancestry” with the actual Bantu-speaking ancestors. They may have been better off saying “BSP-like”. The issue is that the original Bantu speakers were quite genetically similar to their more Westerly neighbors, and for the purpose of the 2017 study, they combined all these related genetic ancestries into one bin. And this broader category accounted for 95.2% of the African DNA in America. Including 99% of the genetic ancestry from Senegambia and Benin, and 97%, from Windward Coast.
As to the Western Central Africa ancestors, their 16% non-Bantu-like DNA was attributed to the rainforest hunter-gatherer populations.
David Marjanović: The more competent/less evil journals precede it by the letter e to make sure everyone understands it’s not a page number…
That sounds like it might be useful, but I’ve never seen it. Interestingly, a number of physics and math journals moved to using article numbers instead of pagination without moving entirely online. For example, the Physical Review journals have used article numbers for a little over twenty years now. Most people certainly read Physical Review articles electronically, but the main journals still have printed versions. (Newer ones, like Physical Review X, are exclusively online.)
@Y:
The real answer is, of course “I don’t know”, but that is dull, so:
Proto-Bantu had an extremely small inventory of consonants, which almost certainly reflects a fairly drastic simplification rather than preservation of some pristine Niger-Congo state. Roger Blench has a paper that we discussed not long ago suggesting that “Bantoid” languages are much more likely to be showing retentions of features that Bantu (a mere twig of the great tree) has lost, than to have contaminated and complicated a Bantu-like purity by association with languages to their West.
But a proper answer would need much more rigorous comparative work than has been achieved to date.
This is just a bit of guesswork on my part, but I have noticed that in the only two Swahili words that I have so far been able to plausibly relate etymologically to Oti-Volta words with labiovelars, (ku)fa “die” and (m)fupa “bone”, the /f/ corresponds to Proto-OV /k͡p/ (Kusaal kpi “die”, Kusaal kɔbir, Moba kpabl “bone”); this is actually not all that mysterious, as it’s the regular outcome of Proto-Bantu /k/ before tense /u/, but it still makes you wonder about the stages prior to Proto-Bantu. Mind you, it’s a line of argument that could cut both ways …
@Dmitry:
Thanks! So they are basically using “Bantu” in a technical sense which only overlaps with the proper linguistic sense to a very limited degree. That seems unhelpful: and liable to lead to actual error given the (alas, well-documented) propensity of perfectly competent geneticists to grossly misunderstand linguistic data. (Still, I suppose we’re in a thought-world where “Caucasian” includes “Welsh” here.)
No, I am being too kind: they need to come up with a better euphemism for what the older ethnographic books call “Negroes” stricto sensu. Calling all such representatives of a genetically defined grouping “Bantu-speaking” (specifically, “speaking”) in defiance of the actual facts about languages like, say, Fon, Twi and Yoruba exemplifies the attitude that “they are all basically the same.” Very regrettable. This shouldn’t be happening in a respectable scientific publication nowadays. And if they don’t even know that these languages aren’t Bantu, they and the peer-reviewers need to be made to read a bloody book before any more trees are sacrificed to produce this sort of stuff.
It is of course stupid in itself to coopt a linguistic term to describe a genetically defined group. One hopes they were not simply prettifying up the Apartheid terminology “Bantu” without thinking it through.
How are you feeling today?
Definitely under the weather — I took a morning nap and an afternoon nap, which is not at all like me, and I’ve got a mild headache — but nothing awful. Thanks for asking!
They are quite a bit more accurate in describing what they measure in the 2017 paper. But by the time it gets to figure captions, it may look unduly blank. They attempted to separate W African from Bantu ancestral components. When the groups are similar in composition, the methods like ADMIXTURE which they use give substantial errors in individual measurements, and averaging mere dozens of people still leaves the errors high. And some groups like Yoruba ended up about 25 percent Bantu which didn’t make much historical sense. Besides, there was no assurance that today’s W Africans are a close representation of the past. So it made good sense to merge these 2 components in the analysis of the mixed Americans, even they had examples of populations in Africa which were just one or just another in composition. It’s a statistical science, nobody wants to claim a conclusion from the fine grades of differences which can’t be reliably tease apart.
It’s a known quandary for the consumer genomic companies. They are under a customer pressure to identify African countries of origin, because their African American customers can’t find where their ancestors came from by the conventional genealogical means. But the African populations are a mosaic of diversity in every region, and identifying there origins even before admixture is a challenge. Normally they are trying to reach a level where 90+% of a region’s population is labeled correctly, and conversely, 90+% of people with the “label” are from the right region. But the white paper of Ancestry.com shows that the standards have to be lowered, sometimes to almost 50%, for the West African regions where slave trade originated. It is a consumer product, not a research paper, so to serve the customer’s needs they need to pick these standards. It is a similar problem with the European Jewish populations and the American descendants lacking any paper trail, just relying on DNA. Some products lump all the ancestral groups, gaining accuracy at the expense of customer satisfaction. Others split and the numbers jump all over. But in a peer reviewed paper it’s always lump.for accuracy..
My opus magnum is PeerJ 6: e5565. (No issue number because no issue.) 199 pages of “paper” plus 191 pages of appendices in “supplementary material” because nobody would have paid for having the whole thing layouted. The PLOS journals also use e.
Interesting indeed. The only journal I’ve seen do that is the Journal of Vertebrate Paleontology, which prints its e-numbers to make sure that taxonomic nomenclatural acts are valid from the date of online publication, not the paper publication many months later, because the International Code of Zoological Nomenclature says this only works if the online version and the paper version are identical in content and layout and doesn’t explicitly state that the page numbers aren’t part of that.
@David Eddyshaw, Dmitry Pruss: This sort of reminds me of the Apartheid South African use of Bantu to describe indigenous South Africans, regardless of whether they were actually Bantu-speaking or not (although, of course, the vast majority were). As the Dutch-descended inhabitants referred to themselves as “Afrikaners,” they wanted another term for the indigenous population.
Incidentally, for cultural reasons which may or may not be totally unrelated, the first name of the anti-Apartheid activist Steve Biko was actually Bantu.
As I may have mentioned already, even I, after just a few years’ residence, could tell with reasonable accuracy whether a Ghanaian was from the north of the county or the south just by looking at them (an unremarkable ability among actual Ghanaians.) As Ghana is only a small part of West Africa, this has naturally left me with a certain scepticism when it comes to the supposed accuracy of genetic studies which purport to show a particular West African genotype.
A (northern) Ghanaian colleague told me that he thought that black Americans looked like northerners rather than southerners on the whole, though I never canvassed opinions on this any more widely, so it may have been just him. Thinking about it, it might have been simply that black Americans usually don’t look particularly like Ashanti, however much they might want (understandably enough) to identify with that charismatic people.
Says Choudhury:
South Eastern Bantu-speakers have a clear linguistic division—they speak more than nine distinct languages
Not by normal linguistic criteria:
WP:
The majority of South Africans speak a language from one of the two principal branches of the Bantu languages that are represented in South Africa: the Sotho–Tswana branch (which includes Southern Sotho, Northern Sotho and Tswana languages officially), or the Nguni branch (which includes Zulu, Xhosa, Swati and Ndebele languages officially). For each of the two groups, the languages within that group are for the most part intelligible to a native speaker of any other language within that group.
Incidentally, “South-Eastern Bantu” doesn’t seem to be an accepted linguistic term. I suspect that it may reflect confusion between ethnicity and language – ironic in the light of the paper’s objectives.
Huh, I hadn’t realized they were that similar. From here:
@David Eddyshaw, it is convenient to say “Europeans” sometimes. Does not mean that Bjork and Sicilians look the same.
We can discuss usefulness of such designations (including “Russian”) but it is not about Africans specifically.
“Bantu”, because the triangle Cameroon-Kenya-Xhosa is very large and because it is a very simple and convenient word.
I think so because I am tempted to use this word myself (and that is when…). That Maasai people (a celebrity, you know) do not speak anything even related is a stronger counter-argument than the West Africa. That Kordofanian languages are Niger-Congo is a counter-argument as well (but only for me).
Compare them to Austronesians.
P.S. I am surprised though that they used it in this context, because exactly West Africa is very famous (again celebrities) in this exact context.
I once had a Zulu boss. A stern looking lady, but not too fierce.
One of the underrated joys of working in a multinational mining corporation – you meet all sorts of people you’d never meet otherwise.
And I hate Apartheid not because they borrowed from linguists a word “Bantu” to designate Bantu-speaking ethnic groups.
“Bantu” instead of “Blacks” or “Negroes”? That was the problem with the regime?
The key thing that’s specific to the South African context is that all the way back to when the Dutch first arrived circa 370 years ago there were two notably different groups of indigenes: a) the various groups now generally lumped together under the 20th-century designator “Khoisan,” who lived among other places right by the Cape of Good Hope; and b) the Bantu-speaking groups farther east, whom the Dutch did not rub up against until they started expanding outward from the immediate vicinity of the Cape. These groups notably differed from each other in physical appearance, culture, and language, and even if they were both “black” they were different sorts of black, such that e.g. an older Afrikaans term roughly synonymous with “Khoisan” is “Kaap Swartes,” to distinguish them from the other sort of “Swartes.” I don’t think apartheid-era government discourse included the remaining “Kaap Swartes” in “Bantu” used as a general classifier, although that was sort of a niche problem since the descendants of those groups had mostly ended up in mixed-race apartheid-system categories like “Cape Coloured” and “Griqua.” (So, e.g. there was no “Bantustan” set up to be nominally ruled by a Khoisan ethnic group, because there wasn’t sufficient numerical critical mass.)
That the Bantu-speaking sort of South African indigenes are more similar (linguistically, culturally, and genetically) to *most* of the rest of the indigenes of sub-Saharan Africa is relevant to how those remaining folks got conceptualized from a South African point of view. If Greenberg had been South African, he might well have just called Niger-Congo “Macro-Bantu” or something like that.
Here by the way is an interesting failed proposal from 1984 to get linguists to stop using “Bantu” (or “Bantoe”) even in purely linguistic contexts, because the word had gotten so skunked under apartheid. https://www.tandfonline.com/doi/abs/10.1080/02572117.1984.10587451
Calling Fon speakers (say) “Bantu speakers” is like calling Welsh speakers “Slavonic speakers.”*
It’s just plain wrong.
Nobody would commit the latter error. You have to ask yourself why they don’t feel the same about the former error. Plain ignorance is the least discreditable reason.
We’re not talking about the blanket use of the term “African”, which may be dull but is not wrong. We’re talking about egregious misuse of a linguistic term in a supposedly scholarly, apparently peer-reviewed scientific publication. No mercy!
*I would be proud to be Slavonic speaker, don’t get me wrong. Nevertheless, such a claim would be fraudulent, alas …
Plain ignorance is the least discreditable reason.
That’s my go-to principle when trying to analyze the behavior of people (including myself). Or rather, it’s the one I end up applying for reasons of charity – after having gone through all the more discreditable explanations, the while chafing and bemoaning. For some reason these always occur to me first.
Charity is more energy-efficient when you can’t change a damn thing anyway. [This kind of cost-benefit analysis has not been considered by utilitarianists, as far as I know. If I’m wrong about that, please cite author and work.]
But charity is disfavored these days — righteous indignation is all the rage.
I know, it’s a downhill battle.
By the way, I got my “jab” on Monday, languished in bed on Tuesday, and today am already racing around again like nobody’s business.
I was so sick and tired of all the squawking that goes on about Covid/vaccinations/who-gets-a-turn etc, couldn’t figure out what was what. Also, two doors down from the supermarket I was headed for, I walked into the doc’s office and asked what’s what. They had a jab “left over”, so I got it. Then I went home unerledigter Dinge because there was a line 20 people long in front of the supermarket. It currently curfew-closes at 21 instead of 22 hours, so apparently these people were panicking …
It must be the Altersbonus. I knew old age must be good for something, if only for jumping the line.
Another thing old age is good for is (generally) less violent reactions to the vaccine.
Nothing to do with language, but for the geneticists among us: Ancient Xinjiang mitogenomes reveal intense admixture with high genetic diversity.
@JWB:
Interesting link. “Kintu” at least has the merit of being logical (though my vote goes to “Luntu” instead.) But “Bantu” is pretty settled as the name of the language group. (Rather like “Eskimo-Aleut.”)
I myself can see no merit whatsoever in using the term “Bantu” as anything other than a linguistic term. Quite apart from its very real potential for offensiveness, using it as a label for ethnic groups is a surefire route to outright factual errors of exactly the sort I have been fulminating about. And “Bantu speaking” is merely a mealy-mouthed way of perpetuating a deep-rooted and fatal methodological error. How about (for comparison) “the Aryan-speaking peoples of Europe”?
Hat: Nothing to do with language, but
David E.: I think part of the problem is that a conceptual scheme that works well enough within one geographical scope or at one level of generality may not work so well in/at a different scope/level. Wikipedia helpfully glosses “Khoisan” as “a catch-all term for the ‘non-Bantu’ indigenous peoples of Southern Africa” and there are practical contexts where it is useful to have a term for the non-Khoisan indigenous peoples of the same region, whether that term be “Bantu” or something else. The problem here, it seems to me, is that you have South African scholars writing on a South African topic for a predominantly South African assumed audience who are along the way mentioning people who live elsewhere on the continent and being quite sloppy and thoughtless in extending South African names-for-categories in a way that doesn’t make much sense in that broader scope/context. And that’s in part because the theoretical mismatch between linguistic categories and ethnic groups and/or genetically-defined populations is sufficiently minor in practice to be overlooked in the South African context but again becomes much more of an issue in the broader context.
I note FWIW that the surnames of Drs. Sengupta and Chaudhury suggest prior membership in yet another racial category that was treated distinctively under the apartheid regime. I don’t know how that might affect their attitudes toward what sort of baggage the word “Bantu” may have in a South African context and what that should mean for usage decisions.
More generally, this is perhaps an instance of the common 21st century phenomenon where a sloppiness or error on a point that the authors’ assumed audience might consider minor or peripheral is on a point not considered minor or peripheral by some of the authors’ actual audience, viz. anyone on the planet with an internet connection who happens to click their mouse in a particular direction.
@J.W. Brewer: The Population Registration Act (1950)* defined Black African and Bantu as synonyms. It is certainly plausible that some of the Khoisan peoples ended up categorized as Coloured or Other; however, I suspect that a lot of them were lumped in with the actual Bantu speakers for Apartheid purposes.
* I happen to remember the name of this law, because it was last of the major Apartheid laws to be repealed, and I remember that there was some fanfare in June 1991 about how its repeal marked the end of the legal basis of the Apartheid system.
authors’ assumed audience …[vs]… actual audience, viz. anyone on the planet with an internet connection who happens to click their mouse
And riding high above them all, pontificating and excoriating, are the influenzers and journalists.
@Brett: Thanks. Apparently when some apartheid legislation was extended to be applicable to Suidwes-Afrika a/k/a Namibia, the legislation so extending it had language like “in such application the word ‘Bantu’ shall be construed as ‘native’ and any compound or derivative of ‘Bantu’ shall be construed as a corresponding compound or derivative of the word ‘native,’” which is consistent with the notion that Khoisan folks who did not get swept into the “Coloured” category were of comparatively de minimis number in South Africa proper but that across the border they were more numerous and politically salient, such that the feeling that they did not fall within the common meaning of “Bantu” was thus harder to ignore and required an explicit statement that the statutory category of “Bantu” was wider than the common meaning.
Yeah, it had occurred to me that the situation was probably somewhat different in South-African-ruled Namibia, but I didn’t know any of the details.
I have been under the impression that the Apartheid regime discerned clearly between Khoi-San and Bantu peoples, seeing the former as unspoiled and noble, even casting itself as the protector of the indigenous peoples against the Bantu invasion. I may have got that impression from the couple of apologetic Wilbur Smith novels I read in the eighties.
I assume that it wasn’t “the regime”, and not even its less strict predecessors. Early in the history of the Cape colony, there was a lot of interracial marriage, both with freed slave girls from Angola / Mozambique / Guinea or Asia and with free local Africans. For some reason they needed to hide or obfuscate the African origin of the local brides, though. In the church books, those were called just “van den Kaap”, Cape’s locals. So centuries later, the genealogists falsely believed that there were virtually no Khoi/San marriages. It took some genetic data, and then critical re-reading of the church books, and then the numbers from both sources matched.
@Dmitry, this is interesting. Do you remember the source?*
I also have always been curious about the history of the name Baasters.
Who used the word originally and what exactly they meant by it.
—
* or any key words that can help me to find it.
De Europese stamvaders van de Basters zijn ‘trekboeren’, kolonisten die met hun vee een zwervend bestaan leiden. Omdat onder deze groep een mannenoverschot bestaat, wordt het aanknopen van seksuele relaties met Khoi-Khoi-vrouwen oogluikend toegelaten.
So the Basters resulted from (tolerated) relationships between pastoral colonists and Khoi-Khoi women, because the former group had too many men.
In 1813 bezoekt LMS-directeur John Campbell Klaarwater. Hij neemt meteen een opmerkelijk besluit: de Basters moeten van naam veranderen. Misschien komt Campbell tot deze beslissing omdat hij verwacht dat het in Engeland moeilijker is om geld in te zamelen voor Bastards dan voor een volk met een meer Afrikaans klinkende naam. Omdat sommige inwoners van Klaarwater denken af te stammen van de Khoi-Khoi-clan Garigurikwa, doopt Campbell de Basters om tot Griekwa’s. Ook hernoemt hij Klaarwater: het dorp staat voortaan bekend als Griekwastad.
So Campbell renamed the Basters to Griekwas, after the Khoi-Khoi clan Garigurikea, perhaps because he thought subscribers in England would not pledge as much money for Bastards. He also renamed Klaarwater to Griekwastad.
Een onvoorzien neveneffect van de naamsverandering is dat de LMS hiermee een nieuwe identiteit creëert die niet op ras of huidskleur is gebaseerd, maar op taal (Kaaps-Hollands) en het christelijk geloof. En dat is uniek in het negentiende-eeuwse Zuid-Afrika. Namen als ‘donkere Hollander’, ‘Hollandse Baster’ en ‘Half-Hottentotse Baster’ verdwijnen langzaam maar zeker uit het taalgebruik.
So the side-effect is that the Griekwa identity is based on language and religion, uniquely for 19C SA. The names with Baster disappear slowly from the language.
Source: https://www.historischnieuwsblad.nl/donkere-hollanders/
Afrikaners are pretty much all “black” according to the one-drop rule.
https://en.wikipedia.org/wiki/Afrikaners#Non-European_ancestry
@de
I like the Native American ancestry referred to in your link. Those Dutch sailors really got around (It was hardly the pastoralists, because the competition for grazing land would not have caused them to build rafts and make the Trek to America). ????.
My note says, “van de Kaap”, and the reference to the preprint form of the publication referenced on wiki, https://www.biorxiv.org/content/10.1101/542761v1
I may have looked at the sources cited within it, too?
@PP – “Asian/Nat Am” is another instance of the statistical genetics’ use of broader ancestry categories (combining several distinct but related ancestry streams into one category) for higher numerical precision, especially when the numbers are small.
Of course DNA is not “liquid like blood” / infinitely missible. It’s passed down from the parents in random fashion in pretty big chunks, averaging 1.5% of the genome in size, and the amount contributed by any specific ancestors starts deviating from the “blood quotients” of 1/4, 1/8 and so on really fast as we go deeper in time. Go a few centuries into the past, and you’d discover that most of your ancestors then didn’t contribute ANY DNA to you. One sees a sizable non-European genetic ancestry in the white South Africans precisely because they had lots of non-white ancestors, rather than someone specific who “contributed” (not really) an infinitesimal droplet of “blood”. It was no secret at all to the South African genealogists that the white families had African and Asian ancestors in any case, because it was in the books of marriages. And as recently as hundred years ago, they continued to marry across the quickly-solidifying official racial lines, but it already required a petition to reclassify the bride into the official “white” category, supported by the parish community. It is visible in the 2019 paper’s data, too, as a segment of the population containing significantly more non-European DNA than average. A friend of mine discovered, after a genetic test, that his grandmother has been officially reclassified as white in order to marry his grandfather. Or rather re-discovered; it turned out that most of the family knew, they were just shy to talk about it. In fact, even he knew that they had Cape Coloured cousins, a rather successful entrepreneurial branch of the family. He just never understood how exactly they were connected to the family tree, and sort of afraid to ask.
But in terms of genetic ancestry, “pure” never means anything other than “mixed deep enough in the past to become a nearly-uniform mix today”, anyway. I shouldn’t be the one to remind you about the differences between a molecular composition and a social construct, right?
The most surprising tidbit from the 2019 paper was that a chunk of chromosome 21 was of Khoi-San origin in 100% of the tested Afrikaners. The only known genes in this region function only in the testes, so perhaps this was a legacy of a “gene drive” in which a selfish gene hijacks meiosis to spread itself at the expense of the rivals, excluding them from the sperm. But I don’t remember ever seeing a followup.
@Dmitry, thank you!
Go a few centuries into the past, and you’d discover that most of your ancestors then didn’t contribute ANY DNA to you.
But the limited size of the population:)
@David E.: the “one-drop rule” in a U.S. context was largely a 20th century concoction, because prior to that the actual historical facts making it unworkable (because it would render too many respectable “white” people non-white if taken literally) were too recent and too widely known. So for example, the South Carolina Constitution of 1895, whose primary purpose was to facilitate the exclusion of blacks from voting in order to solidify the Jim Crow regime, had an anti-miscegenation provision worded as follows: “The marriage of a white person with a negro or mulatto, or person who shall have one-eighth or more negro blood, shall be unlawful and void.” In other words, if your quantum of African ancestry was <12.5% and the rest was white, you were white as a matter of law. Other places in the old South may have at times used a <25% rule.
The <12.5% "safe harbor" was pushed at the convention by former Congressman George Tillman, the less prominent brother of the notorious Pitchfork Ben Tillman, in an oft-quoted speech where he asserted that there was not a single "full-blooded Caucasian" among the "white" delegates at the convention because everyone had some modest amount of "colored" blood.
Didn’t the US at the time have significant immigration from places like Sweden where African “blood” was almost unheard of? You could probably find a Finn or a traveller in most people’s family trees, but Africans didn’t really get there in significant numbers until later.
If you’re referring to Tillman’s speech, there probably weren’t any Swedes at the convention.
Apparently in the full context of Tillman’s remarks he was not referring solely to the most immediately salient sort of “colored blood” viz. “negro,” but also to possible remote ancestry that was Arab or “Mongolian” etc. Really quite forward-thinking compared to your average 1890’s white supremacist in anticipating modern genome studies.
As to the specific issue of white/black admixture in the Old South and the non-applicability of the “one-drop” rule, one study from 2014 (although probably with a suboptimal dataset) found that 12% of self-identified “whites” from South Carolina had at least 1% of African DNA and 5% had 2% or more. As noted above, DNA doesn’t work in neat powers-of-two like family trees do, but it may still be helpful to note that 2% is smaller than 1/32 (= one great-great-great-grandparent) but bigger than 1/64 (= one great-great-great-great grandparent).
Of course the white population of South Carolina these days includes plenty of folks whose ancestors weren’t living there in 1860 or even in 1960 One suspects that if you could limit the analysis to those whose entire family trees were living in the South in 1860 (or >50% of their 1860 ancestors or whatever criterion), the percentage with perceptible African DNA (although still fully consistent with Tillman’s 2% African DNA is lowest tend to be exactly those where the %age of Scandinavian ancestry is highest (Minnesota, the Dakotas, and other states nearby).
I should read the paper, but my backlog is growing. Maybe this weekend.
In other South African news, A Wildfire Has Gutted the University of Cape Town Library and Its Priceless African Studies Collection.
Now I think of it, the genetic heritage of Afrikaners may also explain rooinek (offensive) “English person, English-speaking South African.”
Apparently in the full context of Tillman’s remarks he was not referring solely to the most immediately salient sort of “colored blood” viz. “negro,” but also to possible remote ancestry that was Arab or “Mongolian” etc. Really quite forward-thinking compared to your average 1890’s white supremacist in anticipating modern genome studies.
As to the specific issue of white/black admixture in the Old South and the non-applicability of the “one-drop” rule, one study from 2014 (although probably with a suboptimal dataset) found that 12% of self-identified “whites” from South Carolina had at least 1% of African DNA and 5% had 2% or more. As noted above, DNA doesn’t work in neat powers-of-two like family trees do, but it may still be helpful to note that 2% is smaller than 1/32 (= one great-great-great-grandparent) but bigger than 1/64 (= one great-great-great-great grandparent).
Of course the white population of South Carolina these days includes plenty of folks whose ancestors weren’t living there in 1860 or even in 1960 One suspects that if you could limit the analysis to those whose entire family trees were living in the South in 1860 (or >50% of their 1860 ancestors or whatever criterion), the percentage with perceptible African DNA although still fully consistent with Tillman’s <12.5% rule would be higher than in the not-so-limited sample the study was based on.
Separately, FWIW, the parts of the U.S. where the percentage of self-identified whites with >2% African DNA is lowest tend to be exactly those where the percentage of Scandinavian ancestry is highest (Minnesota, the Dakotas, and other states nearby).
Tillman’s threshold may have been hedging against the bias of familiarity, basically they knew many families with recent slave ancestors, and they knew that this knowledge tends to be suppressed, so they could have assumed that there are lots more cases waiting to come out. It may also be possible that mixed ancestry was more common in the prominent slaveholder families who used abundant slave domestic help, than in the average Southern whites of the era. It is especially likely that the 2014 23andMe’s numbers need to be multiplied by a large factor to account for a relatively small fraction of ancestral African DNA in the victims of the illicit biracial unions. On average, in today’s African American population, the fraction of European ancestry is relatively minor, at about 17%. But in a story after story, the domestics who were abused by their owners turn out to be daughters and granddaughters of other white abusers, sometimes barely a notch above the “Tillman threshold” themselves. These mothers’ “blood” would count as black in a “legal” calculation (I find it hard to use the word legal without some scarequotes here), but they might have contributed a lot less ancestral African DNA than that, maybe 4 or more fold less.
Then there is also a depth-underestimation effect so common in the genealogical tales of anything famous or scandalous in the families. As retold, the most remarkable relatives always appear closer than they really were.
Then there are also deep-seated beliefs about Native American ancestors so common in many old American families.
>Then there are also deep-seated beliefs about Native American ancestors so common in many old American families.
We have one of those family myths. I find it hard to believe my grandmother would have thought it was cool to make this up in the early 1970s. She was not conservative, but she was 65 and not exactly a member of the counterculture either. And much earlier than that, inventing such a story seems even less likely.
My guess is that many of them are true but go back so far that it wouldn’t show up in DNA testing. I’ve tended to believe Elizabeth Warren myself. And also to believe she shouldn’t have marked it on an employment form.
However, I remember doing so on a questionnaire in my teens. Not on anything like a college application. I think it might have been my college athletics media background form. I didn’t play a major sport, and wasn’t a star, so it seemed unlikely anyone would ever see my form. I also made something up to answer the hobby question, and it wound up in a little sidebar in the metropolitan daily paper — Eccentric Hobbies of xxxx Univ. Athletes.
One of the things the 23andMe study indicates (again with whatever caveats about the dataset it had) is that perhaps unlike certain other places like Brazil, some proportions of admixture are much more common than others in the U.S.. So plenty of folks with >60% African DNA, overwhelmingly self-identified as black, and plenty of folks with <10% African DNA, overwhelmingly self-identified as white, but very very few people in for example the 20-35% African DNA range, because the social (and sometimes legal) dynamics influencing who ended up having kids with whom made that not a stable range to stay in over multiple generations. The same study has some info about Native American DNA, which is no more prevalent in self-identified American whites than African DNA (albeit with different regional patterning) although more likely to be talked up and romanticized.
Warren’s story is perfectly true IMVHO yet shows the imprint of the same distortions which I just described. One is the myopic misperception of depth (her actual tribal ancestor appears to have been a generation, or possibly two, more removed on her family tree than the legend went). Another is the markedly reduced fraction of the Native American ancestry in those people who left today’s white ancestors (in Warren’s case, the genetic ancestry of the tribe is reported to me majority-European; in many other legends, the ancestor who was whispered to have been an “Indian” may have been a Mestizo of some fractional level). But Native American ancestry is very hard to analyze at a consumer-genomics level due to the lack of reference samples (it is unethical or illegal to use US Native American DNA for most any research purposes). So at the time of the 2014 23andMe’s study, they were using East Asian DNA as a proxy, and it can markedly underestimate the calculated DNA fractions (because the Ancient North Siberian ancestry stream in the Native Americans is part-shared with the Europeans, and part-overlaps with it, potentially making a part of a person’s Native American genetic ancestry appear as European). Today’s techniques, and some added specimens from Central and South America, can make things markedly more certain, but that’s not how a typical consumer genomics lab does things.
In our lab settings, and in a relatively insensitive genetic ancestry analysis, the self-reported Native Americans and mixed-ancestry Native Americans invariably looked the same as the regular European Americans, sometimes with an African ancestral component, but nothing else. But it could be an artifact of the patient form design. The form asked them to define their ancestry rather than identity. So the only “American” checkbox was “Native American”, and perhaps people who identified as American, and didn’t know anything about their European ancestry, checked this box for lack of a better choice. There are so many Caucasians in this country whose ancestors were all Americans, nary an immigrant. It kind of mirrors the stories of the tribes staying put for 10,000 years. Not to few people checked the Other box to write in White / Caucasian and their variantions….
@Dmitry: About 20 million people, overwhelmingly white, give “American” as their ancestry as the answer to the same open-ended question where you could say “Italian” or “Swedish” or what have you. If you scroll down in this piece you can see a map showing geographical distribution/concentration, which is generally highest in Appalachia and adjacent/similar places. https://en.wikipedia.org/wiki/American_ancestry While some of them may just be giving a none-of-your-business answer to what they consider an intrusive or impertinent question, the usual theory is that most of them descend predominantly from 17th/18th century British incomers (plenty of Scotch-Irish, but not exclusively) who eventually went up into the hills and weren’t the sort of families whose great-aunts kept elaborate genealogies explaining exactly when and from where their forefathers first made landfall on this side of the Atlantic. My family did have that sort of great-aunt, so I think of my ethnic ancestry as English + Scottish + Dutch + Scotch-Irish + German + Huguenot, but maybe “American” would even be a fairer answer, because there was in fact an ethnogenesis on this side of the Atlantic between say 1700 and 1850 as a result of systematic intermarriage among the different sorts of Northwestern-European-Protestant early-ish arrivals.
But if you’re trying to collect ancestry data from a large bunch of white Americans and don’t know about that phenomenon when designing your intake form, you’re asking for trouble.
Elizabeth Warren also did not claim American Indian ancestry on any employment forms. Years after she was hired, she responded to a survey of Harvard faculty asking about whether they had minority ancestry.
asking for trouble
It’s clearly an exaggeration. Even today very few Americans identify as just such in an open-ended ancestry question. 25 years ago, when self-reported ancestry was still the best the geneticists could use, the number was a lot smaller. And in multiple choice questions, fewer still. We fully anticipated that people may not know their Scott from Irish and Dutch from German, so we offered a catch-all Western European ancestry category. And for those who wouldn’t accept any of the boxes, there was “Other”.
The fact that the Native American bin was overwhelmed by people without any obvious Native American genetic ancestry was quite unanticipated though. We don’t know for a fact what prompted people to check it, we just guess. It was tiny fraction of the participants, of course, but there were even fewer genetically-Native Americans among them.
Under the “language spoken at home” question in the American Community Survey, the native American language counts are often much higher than what other sources would indicate; for example, the 2006-2010 tables show 130±53 home-speakers of Penobscot, q.v. (86±42 in an “American Indian area”).
@J. W. Brewer: My ancestry is much like yours. I suspect another factor here is that in the upper South, genealogy tends to be pursued by the dozen families who moved into a given county, engrossed most of the good land, and have spent 200+ years treating everyone else as squatters, and who are always “finding” noble ancestors. The populace in general therefore takes a dim view of the subject. I also suspect that for many of the latter, “American” means “non-monarchist ancestors, unlike those damn genealogy snobs.”
Disclosure: My own ancestry, combining 23andme with some inferences, seems to be 15/16 European, 1/32 African, and 1/32 Native.
“Non-monarchist ancestors,” cousin Rodger? Haven’t the statistical models shown that essentially 100% of living persons of predominantly Western European ancestry are direct descendants of Charlemagne? (And that 100% may have been reached in the European population a sufficient number of centuries back that American blacks (and “Natives”) who have a fairly small percentage of white ancestry from admixture some generations back still essentially all fall into the extended royal family.)
Charlemagne was a mere upstart. True royalty belongs only to the descendents of Childeric.
What about the other descendants of Merovech? Or of Chlodio?
Chlodio, Chlodio, wherefore art thou Chlodio?
David E: Not to the descendants of Vortigern?
We do not speak of Gwrtheyrn, Inviter of the English.
only to the descendents of Childeric
The problem, as far as I can tell from the genealogies (I looked this up a few years ago), is that it seriously looks like either 1) Charlemagne was a descendant of Childeric, or 2) any lines descended from Childeric still around by Charlemagne’s time had died out by the end of his reign.
(It is, of course, easily possible that there were other side-branches that survived past the 8th century but don’t happen to have been sufficiently well recorded.
In addition, there’s some suspicion that Egbert of Wessex, and/or his father Eahlmund of Kent [sic], was actually descended from the previous royal dynasty of Kent [with a well-established Merovingian ancestry – by then the last remaining branch with one], as there were several previous kings in Kent named Egbert but none in Wessex.)
Now I can better understand DP’s complaints about people focusing on male ancestors and Y-chromosomes.
@David Eddyshaw, this is the paper about Bantu speakers via sci-hub. They use “Bantu speakers” correctly.
Thanks, drasvi; the link doesn’t work for me, though. But I’m glad to hear they get it right.
Well, the point was that I browsed through the paper and read most of it. And they are careful enough not to call people to the west of Cameroon Bantu-speakers:)
Looks like my original hope that the fault was with the reporting of the paper* rather than the paper itself is borne out. I’m glad to see the authors vindicated!
Thanks for that, drasvi. (Incidentally, the actual figures about the proportions from West Africa and Central Africa are interesting too. 30% for the latter is higher than I expected.)
*Always a good default assumption, of course.
About sci-hub, sorry, I did not know.
I expected ideological differences (my own position about copyright underwent transformation after a story involving illegal Dostoyevsky in Cairo). I did not knwo that Western countries block it.
In Russia many things are blocked, so opposition-leaning people use VNP or Tor as the default means of browsing the Internet anyway. But they stopped blocking sci-hub: possibly someone told the officials that without it Russian scientists won’t be able to read even works of Russian scientists.
Specifically, the UK blocks the .se sci-hub domain.
I should have thought of that. I’ve managed to access it now, in my capacity as a German. Thanks again, drasvi.
https://xkcd.com/1228/
I mean if you really can’t access neither the paper nor the hub, then you can always ask the LH community. I obviously can’t go to the sci-hub for anything work related, but there are fun things in science which are too far from anything I can get through the regular channels. Just looked up a 1995 malaria paper from Fort Detrick today, in a journal I wouldn’t be able to get anywhere…
Haven’t the statistical models shown that essentially 100% of living persons of predominantly Western European ancestry are direct descendants of Charlemagne?
As I’ve said, it’s a counting argument, not a statistical model except in the broadest sense. Take the number of my geometrically defined ancestors, divide by the population of Europe in the year 800 (of whom perhaps half were his subjects), and the probability that anyone alive then with descendants today is not included many times over is vanishingly small. If I have to give a numerical estimate, I’ll say (2^41)/30,000,000, or 1/73300. (I note in passing that “vanishingly small” is the reciprocal of “astronomically large”.)
Gwrtheyrn, Inviter of the English
He may have been the first but was certainly not the last to bear that title: one might go so far as to say that there was one such in every generation before Edward the Unspeakable.
in my capacity as a German
Do tell!
I suspect he’s referring to the way he fooled the software.
I had thought the .se version had ceased to exist years ago. Maybe it’s just being blocked. There are at least two others that currently work – though Google refuses to find them.
Take the number of my geometrically defined ancestors, divide by the population of Europe in the year 800
and by the same logic, using the worldwide population in 800 CE, you should conclude that Charlemagne is an ancestor of every South-Eastern Bantu and Khoi-San. Since the global population was so much smaller than 2**41 ~~ a couple trillions. So the numeric argument obviously doesn’t work. Generally the parents of the ancestors were much more likely to be members of the same geographical, cultural and class community, with some exceptions (and much bigger exceptions in the mass migration epochs), which for most people means that somewhere between 10 and 20 generations, their trees will begin collapsing, resulting in many of the theoretically possible thousands of ancestors being one and and the same person.
PS: no, the current version is .se, and it is in the US Google edition
Yes, this is how I understood “model”. If one descendant of Charlemagne (or that of his kitchen maid’s lover’s aunt) has visited your small Finnish town in 1232, then you are a proud descendant of Charlemagne, his cook’s lover’s aunt and everyone.
you should conclude that Charlemagne is an ancestor of every South-Eastern Bantu and Khoi-San
Well, no, because Europe formed a single mating pool then, but it did not extend to sub-Saharan Africa.
I’ve read the paper. A couple of notes:
It looks suspicious that the Tsonga and Pedi groups, which contributed by far the most samples to the study, each take up as much of the PC charts as all the smaller groups together.
The geneflow from the Khoi-San into the Southeast Bantu populations is skewed towards females. That’s how it usually goes. More surprising (to me), so is that from Europeans. Patrilocality, but still.
And I want to know more about the first wave of pastoralists, the one that mixed with hunter-gatherers to become Khoekhoe.
a single mating pool then, but it did not extend to sub-Saharan Africa
There is a European admixture in the Sub-Saharan Africa, although minor in scope. But are we sure that Europe can be described as panmictic post-IX c.? Are there some publications on this topic? While it’s well known that Europe is the least genetically varied continent of all, my impression is that the similarities there mostly date back to the great movements of peoples before Charlemagne’s times. Certainly there is a lot of extant genetic structure in Europe, not just on a national basis but small region – to – small region, and some nations exhibit severe founder effects dating back less than a millennium. It sounds to me like something very different from panmixia.
And I want to know more about the first wave of pastoralists, the one that mixed with hunter-gatherers to become Khoekhoe.
The Central Khoisan/Khoe languages are thought to be relatively recent arrivals, according to Güldemann. This would go with their possible relationship to Sandawe (but not Hadza, which is most likely an isolate.)
https://www.researchgate.net/publication/319528342_Khoisan_linguistic_classification_today
I know we’re funning, but of course by “non-monarchist ancestors” I meant “as of 1776.”
David E.: The Central Khoisan/Khoe languages are thought to be relatively recent arrivals, according to Güldemann. This would go with their possible relationship to Sandawe (but not Hadza, which is most likely an isolate.)
[Link to Güldemann 2014]
Thanks. That’s a good summary of the mainstream linguistic consensus — or at least what Güldemann considers to be mainstream consensus — with references to forthcoming papers that we may hope has since forthcome. 2014 was probably too early for cross-desciplinary use of archaeological and archaeo-genetic evidence, but I see that he was a co-author on:
Vladimir Bajić et al: Genetic structure and sex‐biased gene flow in the history of southern African populations, AJPA, 2018
I haven’t read it yet, but suoerficially it seems to have more to say about the Bantu expansion than the older Khoe-Kwadi wave, which is probably also more straightforward when it’s based on modern DNA and not archaeo-genetics.
I’ve read somewhere that pastoralism arrived in South Africa 2000 years ago from Eastern Africa.
Presumably the people who brought it spoke Nilo-Saharan languages, which were erased by subsequent Bantu expansion.
Then it is a material for alternative history.
Instead of Bantu expansion to SA there has been extensive exachange with the east coast. By the time Europeans arrive they find medieval kingdoms, populated by speakers of click langauges and proud carriers of steatopygia, possibly operating printing presses. Or maybe those arrive to Europe first.
I wish there was more alternative history like that instead of the usual boring Western European/American variations.
@Trond:
There’s a good bit on genetics in the book that chapter is taken from.
@SFReader:
IIRC the idea is that it was Khoe speakers who introduced pastoralism, probably not too many centuries before Bantu speakers arrived with their fancy agriculture and iron-mongering. You don’t need to speak a Nilo-Saharan language to be a pastoralist, after all, although no doubt it helps …
(Also, there is no Nilo-Saharan family. It is not the spoon that bends. It is only yourself.)
proud carriers of steatopygia, possibly operating printing presses.
What a wonderful image ! Fat buns are natural calenders.
This brings to mind the core function of photocopiers at office parties.
pastoralism arrived in South Africa 2000 years ago from Eastern Africa.
It’s relatively hard to know, at this point in time, where and when the interactions between various streams of ancestry contributing to the present-day genetic ancestry of the Khoi and San peoples took place. Their type of ancestry was much more widespread in the past, found in the areas of today’s Malawi and Tanzania as recently as early centuries CE in the ancient-DNA studies of Skoglund 2017, Schlebusch 2017. On the other hand the San-like ancestry was already present in South Africa 2,000 years ago. Of course we seem to believe, without conclusive data, that the hunters-gatherers of Africa lived in their ancestral places for eons, and accepted genetic admixtures in situ, even when we know from both linguistic / historic and ancient DNA studies in the Northern Eurasia and the Americas that the hunters-gatherers have been extremely mobile and the population turnover was rampant. But in actuality we often need to decide between the 3 different scenarios of admixture process: a) a direct interaction in situ (the source group migrated in the past) b) a direct interaction followed by migration of the target group, and c) indirect interaction (say a wave of the migrating Bantu pastoralists picked up a more ancient East African ancestry and passed it on as it migrated further South). Ultimately, more ancient DNA samples may shed better light on what happened where.
Some timing issues are easier to resolve with haplotype-length studies in the autosomal DNA, and with phylogenies. For example, Naidoo 2020 analyzes Y phylogenies and concludes that there were two streams of E haplogroups into the Khoi-San, one with the Bantus and one from the East African pastoralists who predated them (and are assumed to have migrated to South Africa ~ 2KyA), plus an additional, even earlier possibly East-South flow detected only in the San (the closest kin in this B-haplogroup branch were the Hadza, but the common roots are too deep, 20,000 years ago; nevertheless, there is a shared autosomal ancestry between the San and the Hadza too, which makes the authors stick to the Eastern-origin hypothesis)
A more recent paper on the contemporary Khoi-San genomes (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7530619/ ) hypothesizes multiple East African gene flows; as always, genomic techniques are more suitable for just one-time admixture scenarios than for telling apart numerous admixture events from related populations.
as always, genomic techniques are more suitable for just one-time admixture scenarios than for telling apart numerous admixture events from related populations.
If I understand this correctly (by no means certain), it implies that “genomic techniques” are worthless for distinguishing “numerous admixture events from related populations”. If one had independent reasons to postulate “numerous admixture events”, what techniques would used to distinguish them, if not “genomic techniques” as well ?
I’m trying to follow the form of the argument here, aka identify the syllogism pattern being used. That’s why I put all those expressions in quotes.
I probably don’t understand the question. Are you asking about the semantic differences between “less suitable”, “harder to use”, and “worthless”? Or about explaining how admixture-dating works, and what are the difficulties there? I can try the latter, but I am not sure if this is what’s asked.
But basically an ancestor’s DNA gets gets broken down to smaller and smaller chunks with each passing generation. If two groups of your ancestors were genetically distinct, then these chunks will have slightly distinct content (it’s oft repeated that all humans are 99% similar by their DNA content, so the telltale differences will be few). The less distinct were the ancestral peoples, the longer a chunk of DNA one might need to tell from whom it originated. The harder it will be to tell where a DNA sequence from one population ends, and from the other begins. But the statistical distribution of the ancestral fragments’ length is the main indicator of how many generations have passed since the mixing event. So imprecision in DNA segment lengths gradually degrades precision of age-of-admixture.
There are good statistical algorithms to “digest” the aggregate data, despite all the random deviations, and to estimate the most likely admixture date. If there were multiple admixtures, then the longest chunks of ancestral DNA turn out to be the most informative, and so the estimated most likely admixture date is generally close to the most recent of the major admixture events. One can iterate further, asking if any admixing happened earlier, and left relatively smaller, less abundant, less precisely measured DNA segments. But it yields progressively wider confidence intervals, and, if the ancestries were too close to start from, then the intervals may be too wide to be useful. But it is not unsurmountable. Having lots more genomes sequenced improves statistical accuracy. Having ancient genomes sequenced (when the admixture dates were much closer, and left much longer chunks of DNA) also improves statistical accuracy.
Both developments are happening now. Say just published Bacho Kiro ancient genomes clearly show multiple minor waves of Neanderthal admixtures, while today’s European DNA can see more or less just one with any precision. Or, also just published Oceania genomes, just by sheer amount of sequenced modern DNA, clearly date multiple waves of Denisovan ancestry. These are sort of extreme examples, with tens of thousand years separating us from those events, but probably a good illustration of how the continuing progress makes predictions more powerful.
It is not the spoon that bends. It is only yourself.
My spoons bent so much that I had to go out and buy half a dozen bending-resistant spoons, none of which, fortunately, have caught bending disease from their drawermates. (Which is not to say that the bendables are without their uses, e.g. eating yogurt.)
I get my bend-resistant spoons for free at fast-food places. They are plastic, and they break instead of bending.
N.B. Uri Geller came once to our house when I was a kid, and by my request demonstrated some magic. I was not impressed.
drasvi: Instead of Bantu expansion to SA there has been extensive exachange with the east coast. By the time Europeans arrive they find medieval kingdoms, populated by speakers of click langauges and proud carriers of steatopygia, possibly operating printing presses. Or maybe those arrive to Europe first.
In this timeline a fascination with African style in the 19th century made the bustle popular in Europe.
David E.: There’s a good bit on genetics in the book that chapter is taken from.
I’ll go looking,
Dimitry: A more recent paper on the contemporary Khoi-San genomes
Thanks. Skoglund 2017 is much cited. It identifies several distinct ancestral populations, but the data isn’t fine-grained enough to disentangle their paths into southern Africa.
And there are interesting stories behind it. The spread of livestock breeding must have brought language and genes with it all the way from the Levant to the Cape, but apparently not the same language family and the same genetic profile all the way.
Roget Blench has notably suggested about a decade ago that Khoe pastoralism was adopted from Southern Cushitic, proposing also that a word for ‘cow’ found in e.g. Nama koma (and further afield e.g. !Xóõ gùmi) originates from Cushitic, citing e.g. Proto-Agaw “*kəmi”, Proto-East Cushitic “*korma” ‘bull’. So pre-Bantu, but still not any flavor of “Nilo-Saharan” either.
I have no idea where these reconstructions come from, though… Some checking of the Moscow School databases does turn up a Proto-Agaw *kim ‘cow’ (> e.g. Qimant /kemaː/, Awngi /kemiː/, Bilin /kim/), but no other proposed cognates in Cushitic and also nothing even vaguely resembling *korma. The common SCushitic terms seem to be instead /aw/ ‘bull’, /ɬeː/ (? *ɬaHe) ‘cow’.
Oliveira, Fehn, Anorim, Stoneking, Rocha: Genome wide variation in the Angolan Namib desert reveals unique Pre-Bantu ancestry
Some background of linguistic relevance:
The “peripatetic groups” are populations that become economically dependent on but not fully integrated in another and economcally stronger population. In the Namib desert it leads to a caste-like structure.
Several groups became peripatetic to Bantu pastoralist societies. They all ended up taking up the dominant language, and what’s more, having a significant part of their genes from the majority population — a genetic one-way stream.
One “peripatetic group”, the Damara from Namibia, is instead thought to be an originally Bantu group that became peripatetic to the incoming Nama pastoralists and picked up a Khoi-Kwadi language frm them, It’s not dscussed in the article, but the genetic one-way stream from the dominant group doesn’t seem to be mirrored here.
The overarching ethno-lingustic history seems to be:
1. There were two genetically distinct populations of foragers in southern Africa, one “southern” and one “southwestern”.
2. The first wave of pastoralists came from East Africa and mixed with the southern foragers. All the extant “Khoi-San” peoples are from this pool.The southwestern foragers were largely unaffected.
3. The Eastern and Western Bantu met and mixed with very different populations. Bantu peoples in Angola and Namibia are heavily mixed with southwestern foragers, those in South Africa and Botswana with the southern Pre-Bantu peoples.
4. The societies of Pre-Bantu foragers and herders developed different strategies. Some southern groups migrated north, some became “peripatetic” to economically stronger groups and eventually lost their language.
I’d guess that also the Southern Bantu societies might contain undetected “peripatetic” groups of Pre-Bantu origin, but I don’t know enough about the traditional social structures to say.
I get my bend-resistant spoons for free at fast-food places. They are plastic, and they break instead of bending.
If I want to serve ice cream (e.g.) I need a bend-resistant spoon. One that bends plastically (metal) and one that bends elastically and then breaks (plastic) are equally useless for the purpose.
I recommend a titanium spork.
Stainless steel is fine, if sufficiently thick.
Another interesting paper about the remnants of the pre-Bantu hunters-gatherers among the marginalized groups which adopted the languages of their Bantu agriculturalist neighbors, but largely retained their original lifestyles. (They may be retaining elements of their original languages too, as we discussed before with regard to the surprise discoveries that the marginalized herders of Southern Angola retained their distinct, Khoe-San related language which was presumed extinct).
The new paper is about the grasslands of Zambia, where agriculture was first brought in ~100 CE and the Bantu cultures started dominating since ~400 CE, but distinct hunters-gatherers survived on the fringes well into XIX c., especially in the areas where tsetse fly limited the spread of agriculture. The paper focuses on BaTwa, “the strangers”, an umbrella term for marginalized wetland and floodplain dwellers who often specialize in fishing. BaTwa speak various languages of their neighbors. Their very identity is stigmatized and it isn’t easy to probe their folklore and linguistics, but both their neighbors and themselves concur that the ancestors of the BaTwa were already there when their current agricultural neighbors moved in. The study shows that they retain substantial hunter-gatherer ancestry which is distinct from either Khoe-San or from Congo Rainforest hunters-gatherers (and may be from the source intermediate between these two). A very low level of the same previously unknown HG ancestry is also apparent in Zambian agriculturalists.
https://www.nature.com/articles/s41467-024-50733-y
The Zambian BaTwa groups have similar genetic makeups as the Zulu and Xhosa of South Africa — a near-complete dominance of “Bantu” Y chromosomes and a more even balance between local and Bantu women. That looks more like frontier communities established by incoming men and local women, followed by family immigration or intermarriage with off-frontier communities.